间接自适应动态递归模糊神经网络控制器设计

来源期刊：中南大学学报(自然科学版)2004年第2期

论文作者：张友旺 王荣铸

文章页码：253 - 257

关键词：动态递归;模糊神经网络;动态非线性系统;自适应控制;投影算法

Key words：dynamic recurrence; fuzzy network; dynamic nonlinear system; adaptive control; projection algorithm

摘    要：针对仿射非线性系统,提出了一种新型的基于动态递归模糊神经网络(DRFNN)的间接自适应控制器。该控制器采用DRFNN对系统的动态非线性映射进行在线估计,并依据李亚普诺夫稳定性理论推导出DRFNN参数在线调整的自适应算法,同时运用投影算法确保参数向量处于约束集合内。应用自适应DRFNN对动态非线性映射进行在线估计时,仅采用被控系统的1个状态变量作为其输入,避免了因增加输入个数而导致网络结构膨胀的问题,从而加快了收敛速度。仿真结果表明:由自适应DRFNN构成的控制器可使系统具有满意的跟踪性能。

Abstract: A novel indirect adaptive controller based on dynamic recurrent fuzzy neural network(DRFNN) is proposed for affine nonlinear system. In this controller, DRFNN is adopted to evaluate the dynamic nonlinear map online. The online adaptive algorithm for the parameters of DRFNN is formulated according to the Lyapunov stability theory, and the projection algorithm is adopted to keep the adapted parameters within the limited sets. Only one state variable of the controlled system is used as its input while evaluating the dynamic nonlinear map online, so the structure expansion of the networks resulting from the increment of the number of its inputs can be avoided, and the convergence can be speed up accordingly. The simulation results show that the controller constructed by adaptive DRFNN can make the controlled system satisfying the tracking performance.